DataBench: leaderboard
Real-world tabular question-answering benchmark over many datasets, used in SemEval 2025 Task 8.
Metric: Score (%). Source: huggingface.co. Source: hex.tech. Saturation forecast: Around February 2028. 72 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max) | 70.5 | #3 (Claude Opus 5.5) |
| 2 | Claude Opus 5.5 (xHigh) | 67 | #3 (Claude Opus 5.5) |
| 3 | Claude Opus 5.5 (High) | 65 | #3 (Claude Opus 5.5) |
| 4 | Claude Opus 5.5 (Medium) | 62.5 | #3 (Claude Opus 5.5) |
| 5 | GPT-6 Sol (xHigh) | 61.3 | #14 (GPT-6 Sol) |
| 6 | GPT-6.1 Sol (Medium) | 58 | #7 (GPT-6.1 Sol) |
| 7 | GPT-6.1 Sol (Max) | 57 | #7 (GPT-6.1 Sol) |
| 8 | GPT-6 Sol (Max) | 57 | #14 (GPT-6 Sol) |
| 9 | GPT-6 Astra (Max) | 55 | #6 (GPT-6 Astra) |
| 10 | GPT-6 Astra (Low) | 55 | #6 (GPT-6 Astra) |
| 11 | GPT-6.1 Sol (xHigh) | 55 | #7 (GPT-6.1 Sol) |
| 12 | GPT-6.1 Sol (High) | 55 | #7 (GPT-6.1 Sol) |
| 13 | GPT-6 Sol (High) | 54.7 | #14 (GPT-6 Sol) |
| 14 | Claude Opus 5.5 (Low) | 54 | #3 (Claude Opus 5.5) |
| 15 | GPT-6 Astra (Medium) | 52.7 | #6 (GPT-6 Astra) |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=databench · How It Works · Data refreshed daily, snapshot 2026-10-20.